The Pattern recognition framework and Hoeffding's bounds

نویسنده

  • Yoav Freund
چکیده

One of the main restrictions of the PAC framework that makes it unrealistic is that the labels of the instances are assumed to be generated by a concept from a predetermined concept class. In real world applications of learning we can rarely assume such prior knowledge. A more realistic framework of analysis is the “Pattern recognition problem” framework proposed by Vapnik [5]. In this framework we make the much weaker assumption that examples are generated independently at random from some unknown distribution and that we have reason to believe that some concept from a know concept class is likely to be a good predictor of the label given the instance for this distribution. More formally, we define the interaction between the teacher and the student as follows (Compare this to the student-teacher interaction defined for the PAC framework).

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A novel grey–fuzzy–Markov and pattern recognition model for industrial accident forecasting

Industrial forecasting is a top-echelon research domain, which has over the past several years experienced highly provocative research discussions. The scope of this research domain continues to expand due to the continuous knowledge ignition motivated by scholars in the area. So, more intelligent and intellectual contributions on current research issues in the accident domain will potentially ...

متن کامل

Classifier Ensemble Framework: a Diversity Based Approach

Pattern recognition systems are widely used in a host of different fields. Due to some reasons such as lack of knowledge about a method based on which the best classifier is detected for any arbitrary problem, and thanks to significant improvement in accuracy, researchers turn to ensemble methods in almost every task of pattern recognition. Classification as a major task in pattern recognition,...

متن کامل

Object Recognition and Performance Bounds

Abst rac t . Object recognition is the classification of objects into one of many a p m o r i known object classes. In addition, it may involve the estimation of the pose of tile object and/or the track of the object in a sequence of images. Bayesian statistical pattern recognition, neural networks and rule based syst~ems have been used to address the object recognition problem. In the case of ...

متن کامل

Upper and Lower Tight Error Bounds for Feature Omission with an Extension to Context Reduction

In this work, fundamental analytic results in the form of error bounds are presented that quantify the effect of feature omission and selection for pattern classification in general, as well as the effect of context reduction in string classification, like automatic speech recognition, printed/handwritten character recognition, or statistical machine translation. A general simulation framework ...

متن کامل

Iterative Weighted Non-smooth Non-negative Matrix Factorization for Face Recognition

Non-negative Matrix Factorization (NMF) is a part-based image representation method. It comes from the intuitive idea that entire face image can be constructed by combining several parts. In this paper, we propose a framework for face recognition by finding localized, part-based representations, denoted “Iterative weighted non-smooth non-negative matrix factorization” (IWNS-NMF). A new cost fun...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2004